Received-SPF: pass (google.com: domain of linux-kernel+bounces-20389-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Tue, 9 Jan 2024 10:34:54 +0530
From: Pavan Kondeti <quic_pkondeti@quicinc.com>
To: Mukesh Ojha <quic_mojha@quicinc.com>
CC: <andersson@kernel.org>, <konrad.dybcio@linaro.org>,
        <linus.walleij@linaro.org>, <linux-arm-msm@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, <linux-gpio@vger.kernel.org>
Subject: Re: [PATCH v11 1/4] firmware: qcom: scm: provide a read-modify-write
 function
Message-ID: <6e99c4ff-a377-4385-b79c-60438e3e8735@quicinc.com>
References: <1704727654-13999-1-git-send-email-quic_mojha@quicinc.com>
 <1704727654-13999-2-git-send-email-quic_mojha@quicinc.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <1704727654-13999-2-git-send-email-quic_mojha@quicinc.com>

On Mon, Jan 08, 2024 at 08:57:31PM +0530, Mukesh Ojha wrote:
> It was realized by Srinivas K. that there is a need of
> read-modify-write scm exported function so that it can
> be used by multiple clients.
> 
> Let's introduce qcom_scm_io_rmw() which masks out the bits
> and write the passed value to that bit-offset.
> 
> Suggested-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
> Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
> Tested-by: Kathiravan Thirumoorthy <quic_kathirav@quicinc.com> # IPQ9574 and IPQ5332
> ---
>  drivers/firmware/qcom/qcom_scm.c       | 26 ++++++++++++++++++++++++++
>  include/linux/firmware/qcom/qcom_scm.h |  1 +
>  2 files changed, 27 insertions(+)
> 
> diff --git a/drivers/firmware/qcom/qcom_scm.c b/drivers/firmware/qcom/qcom_scm.c
> index 520de9b5633a..25549178a30f 100644
> --- a/drivers/firmware/qcom/qcom_scm.c
> +++ b/drivers/firmware/qcom/qcom_scm.c
> @@ -19,6 +19,7 @@
>  #include <linux/of_irq.h>
>  #include <linux/of_platform.h>
>  #include <linux/platform_device.h>
> +#include <linux/spinlock.h>
>  #include <linux/reset-controller.h>
>  #include <linux/types.h>
>  
> @@ -41,6 +42,8 @@ struct qcom_scm {
>  	int scm_vote_count;
>  
>  	u64 dload_mode_addr;
> +	/* Atomic context only */
> +	spinlock_t lock;

IMHO, this comment can be confusing later. one might think that
qcom_scm_call_atomic() needs to be called with this lock, but that does
not seems to be the intention here.

>  };
>  
>  struct qcom_scm_current_perm_info {
> @@ -481,6 +484,28 @@ static int qcom_scm_disable_sdi(void)
>  	return ret ? : res.result[0];
>  }
>  
> +int qcom_scm_io_rmw(phys_addr_t addr, unsigned int mask, unsigned int val)
> +{
> +	unsigned int old, new;
> +	int ret;
> +
> +	if (!__scm)
> +		return -EINVAL;
> +
> +	spin_lock(&__scm->lock);

So, this function can't be called from hardirq context. If that ever
happens, with this new spinlock (without disabling interrupts), can
result in deadlock.

> +	ret = qcom_scm_io_readl(addr, &old);
> +	if (ret)
> +		goto unlock;
> +
> +	new = (old & ~mask) | (val & mask);
> +
> +	ret = qcom_scm_io_writel(addr, new);
> +unlock:
> +	spin_unlock(&__scm->lock);
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(qcom_scm_io_rmw);

Thanks,
Pavan